RELATIONSHIPS AMONG CRUDE PRICE, GOLD PRICE AND STOCK
MARKET-AN EVIDENCE OF BOMBAY STOCK EXCHANGE
Dr. Amalendu Bhunia
Associate Professor, Dept. of Commerce
University of Kalyani, West Bengal, India
bhunia.amalendu@gmail.com
RELATIONSHIPS AMONG CRUDE PRICE, GOLD PRICE AND STOCK
MARKET-AN EVIDENCE OF BOMBAY STOCK EXCHANGE
Abstract
The present paper explores the relationships among crude oil price, gold price and stock
market in India. Increasing crude oil prices will increase the cost of production that will
affect cash flow and will decrease stock prices. Indian investors are demonstrating uncase
in the stock markets due to continuous rising of gold prices on account of no fear and no
future loss. However, this study is based on secondary data obtained from various data
sources including BSE database and World Gold Council database for the period from
January 2, 1991 to October 31, 2012. In the course of analysis, Johansen cointegration
analysis and Granger causality test have been designed. Johansen cointegration test result
indicates that there exists a long-term relationship among the selected variables. Granger
causality test result shows that there must be either bidirectional or no causality among
the variables.
Keywords: Crude oil price, gold price, sensex, Johansen cointegration test, Granger
causality test
COINTEGRATION AND CAUSALITY RELATIONSHIP AMONG CRUDE
PRICE, DOMESTIC GOLD PRICE AND STOCK MARKET-AN EVIDENCE OF
BOMBAY STOCK EXCHANGE
1. INTRODUCTION
Crude Oil and gold are the two tactical commodities that have acknowledged much
interest in recent times in India because of their rising flows in prices as well as economic
uses (Le Thai-Ha et al, 2011). Increasing crude oil prices will increase the production
costs which will affect cash flow and will decrease stock prices (Ayhan Kapusuzoglu,
2011). The global economic turmoil is likely to stimulate uncertainty in gold prices,
which has already made it a risky asset for investors in India. Many researchers have
been done the long-run and short-run relationships among stock price index and gold
price in developed and developing countries. Empirical results show that gold price can
greatly affect the stock market (Mahmood Yahyazadehfar and Ahmad Babaie, 2012).
Accordingly, in the last two decades, stock markets in India have received a great deal of
concentration, both as a source of financial development and finally economic growth,
and in the context of large swings in stock market valuation (Bose, 2005; taken from
Ranjan Dasgupta, 2012). In view of this, this paper aims to examine the cointegration and
causal relationships between crude oil price, domestic gold price and stock market in
India. Section two briefly peep the review of related literatures across the world. In
section three, the materials and methods adopted are presented. The empirical results and
inferences are discussed in section four and section five concludes the paper.
2. REVIEW OF LITERATURES
Ranjan Dasgupta (2012) has made a study to examine the long-run and short-run
relationships between BSE SENSEX and four key macroeconomic variables (wholesale
price index, index of industrial production, exchange rate and call money rate) of Indian
economy by using monthly data from April, 2007 to March, 2012 with the application of
financial econometrics. Empirical results of the study showed that there are no short-run
causal relationships between SENSEX and four macro-economic variables but confirmed
long-run relationships between BSE SENSEX with index of industrial production and
call money rate.
Subarna K. Samanta and Ali H. M. Zadeh (2012) examined the co-movements of selected
macro-variables (gold price, stock price, real exchange rate and the crude oil price) based
on 21 years data using econometric models for the periods from January 1989 to
September 2009. The study exposes that there is a cointegration relationship between the
variables.
S. Kaliyamoorthy and S. Parithi (2012) have made a study to examine the relationship
between gold price and stock market for the period from June 2009 to June 2010. They
prove that there is no relationship with the stock market and gold price and stock market
is not a ground for rising gold price.
Le Thai-Ha et al (2011) have made a study to investigate the relationships between the
prices of two strategic commodities, that is, gold and oil in terms of index of US dollar by
using monthly data from January, 1986 to April, 2011 with the application of financial
econometrics. Empirical results of the study showed that there is a long-run relationship
existing between the prices of oil and gold and the oil price can be used to predict the
gold price.
Seyed Mehdi Hosseini et al (2011) examined the relationships between stock market
indices and four macroeconomics variables, namely crude oil price, money supply,
industrial production and inflation rate in China and India using yearly data between
January 1999 and January 2009. The study point out that there are both long and short
run linkages between macroeconomic variable and stock market index in each of these
two countries.
Gagan Deep Sharma and Mandeep Mahendra (2010) made a study to evaluate the
long-term relationship between BSE and Macro-economic variables (exchange rates,
foreign exchange reserve, inflation rate and gold price) for the period from January 2008
to January 2009 using multiple regression model. The study reveals that exchange rate
and gold price influences the stock prices in India.
The conclusive sum of this retrospective review of relevant literature produced till
date on the offered subject reveals wide room for the soundness and instigates of this
work and replicates a few crucial supports that assert its feasibility, as may be marked
here it. The subsistence of crude oil price, gold price and stock price indices of stock
market in India are hardly available. Therefore, the present study aspires to observe the
changes or increase in daily crude oil price, gold price and its impact on sensex in India.
3. MATERIALS AND METHODS
3.1 Data source
The study is based enormously on secondary data acquired from different data sources
including BSE database and the database of world gold council for the period from
January 2, 1991 to October 31, 2012.
3.2 Sample design
This study considers daily data encircling the closing stock price indices of BSE
(SENSEX), the daily gold price indices and daily crude oil indices (WCI). After
appropriate fitting the daily closing indexes with the resultant gold price and crude oil
price, there are 5274 observations. Eviews 7.0 package program has been used for
arranging the data and execution of econometric analyses.
3.3 Tools used
In the course of analysis of the present study, only econometric tools include Augmented
Dickey Fuller (ADF) both at levels and 1st differences, Johansen’s system co-integration
test and Granger causality test have been used.
3.4 Model specification
Unit root test
A time series volatile data is stationary or not or contain unit root for which Augmented
Dickey-Fuller (ADF-1979) test method has been used in the present study. The time
series data is non-stationary if the critical value is lower than the calculated value and
afterwards null hypothesis is rejected and series is decided to be stationary.
H0: Series is stationary
H1: Series is non-stationary
If all the sets of data are found I (1) (non-stationary), and if the regression produces a I
(0) error term, the equation is said to be co-integrated. On the other, if there are two
variables, xt and yt, which are both non-stationary in levels but stationary in first
differences, then xt and yt would become integrated of order one, I(1), and their linear
combination should have the form:
zt = xt - ayt
(1)
[Claire G. Gilmore et al, (2009)]
However, if there is a I (0) such that zt is also integrated of order zero, I (0), the linear
combination of xt and yt is said to be stationary and the selected variables are also
to be co-integrated (Engle & Granger, 1987). If two variables are co-integrated,
there will be a contributory long-run relationship between them.
On the other hand, for determining the charisma of unit roots, an expansion of the
Dickey and Fuller (1981) method has been applied. The ADF test uses a regression of
the first differences of the series against the series lagged once, and lagged difference
terms, with optional constant and time trend terms:
∆yt = a0 + a1t + γyt-1 + Σbiyt-1 + et
(2)
In the equation ∆ is the first-difference operator, a0 is an intercept, a1t is a linear time trend,
et is an error term, and i is the number of lagged first-differenced terms such that et is
the white noise. The test for a unit root has the null hypothesis that signifies γ = 0.
If the coefficient is significantly different from zero, the hypothesis that yt encloses a unit
root is considered as rejected. If the test on the level series fails to reject, the ADF
procedure is then applied to the first-differences of the series. Rejection escorts to the
finale that the series is integrated of order one, I (1).
A drawback of the Dickey-Fuller test is its assumption that the errors are statistically
independent and have constant variances.
Johansen co-integration test
Co-integration tests endow with a mean to establish whether a set of endogenous
variables share a common long-run stochastic trend. A pronouncement of co-integration
specifies interdependence of the endogenous variables, which may be the result of
economic connections between the markets or arbitrage activity between investors.
Hypothesis to be examined with Johansen co-integration test to be applied on the study
has been presented below:
H0: There is no co-integration relationship between variables
H1: There is co-integration relationship between variables
The Johansen (1988) approach to testing for co-integration relies on the relationship
between the rank of a matrix and its characteristic roots, or eigenvalues. Let Xt be a vector
of n time series variables, each of which is integrated of order (1), and assume that Xt can
be modelled by a Vector Auto Regression (VAR):
Xt = A1xt-1 + ... + Apxt-p + εt
(3)
Rewriting the VAR as
∆xt = ΠXt-1 + ΣΓ∆Xt-i + εt
(4)
Where, Π = ΣAi - I, Γi = - ΣAi. If the coefficient matrix Π has a reduced rank r < k,
there exists k x r matrices α and β each with rank r such that Π = αβ´ and β´x t are
stationary. The number of co-integrating relations is given by r, and each column of β is a
co-integrating vector. Equation (4) can be modified to allow for an intercept and a linear
trend.
The number of distinct co-integrating vectors can be obtained by determining the
significance of the characteristic roots of Π. To identify the number of characteristic
roots that are not different from unity we have used two statistics, the trace test and the
maximum eigenvalue test:
λtrace(r) = -TΣln(1 - λi)
(5)
and
λmax (r,r+1) = -Tln(1 - λr+1)
(6)
Where, λi = the estimated values of the characteristic roots (eigenvalues) obtained from
the estimated Π matrix, r is the number of co-integrating vectors, and T = the number of
usable observations. The trace test evaluates the null hypothesis that the number of
distinct co-integrating vectors is less than or equal to r against a general alternative
hypothesis (the number of distinct co-integrating vectors is more than or equal to r).
The maximum eigenvalue test examines the number of co-integrating vectors versus that
number plus one. If the variables in Xt are not co-integrated, the rank of Π is zero and
all the characteristic roots are zero. Since ln (1) = 0, each of the expressions ln (1 - λi) will
equal zero in that case. Critical values for the test are provided by Johansen and
Juselius (1990) and by Osterwald-Lenum (1992).
Pairwise Granger causality Tests
We test for the dearth of Granger causality by estimating the following VAR model
(Olushina Olawale Awe, 2012):
Yt = a0 + a1Yt-1+…+ apYt-p+ b1Xt-1+…+ bpXt-p+Ut
(7)
Xt = c0 + c1Xt-1+…+ cpXt-p+ d1Yt-1+…+ dpYt-p+Vt
(8)
Testing
H0:b1=b2=…=bp=0 against H1: Not H0 is a test that Xt does not Granger-cause Yt.
Similarly, testing H0: d1= d2=…= dp=0 against H1: Not H0 is a test that Yt does not
Granger cause Xt.
In case of Granger causality between the two variables, null hypothesis is rejected if the
probability value is less than alpha (0.05).
4. EMPIRICAL RESULTS AND ANALYSIS
4.1 Unit Root Test Results
Johansen cointegration analysis is needed where there are any long-term cointegration
relationships between crude oil price, gold price, exchange rates and stock price indices
of BSE and NSE. Cointegration analysis is possible if the series are stationary. In order
to stationarity analysis, unit root test of Augmented Dickey-Fuller (ADF) is conducted
with the levels and first differences of each series on the condition that the null
hypothesis is non-stationary, so rejection of the unit root hypothesis supports stationarity.
Table 1: Results of Augmented Dickey-Fuller Test at level
Intercept but no trend
Intercept and trend
Variables
Test
Critical Prob.
Test
Critical
Prob.
statistics value
statistics
value
(1%)
(1%)
Gold price
-1.34
-3.43
0.9998
-1.14
-3.96
0.8999
Sensex
-0.57
-3.43
0.8746
-2.05
-3.96
0.5714
Crude price
-0.53
-3.43
0.8831
-2.44
-3.96
0.3586
Table 2: Results of Augmented Dickey-Fuller Test at 1st difference
Intercept but no trend
Intercept and trend
Variables
Test
Critical Prob.
Test
Critical
Prob.
statistics value
statistics
value
(1%)
(1%)
Gold price
-20.69
-3.43
0.0000
-20.69
-3.96
0.0000
Sensex
-66.99
-3.43
0.0001
-66.99
-3.96
0.0000
Crude price
-73.69
-3.43
0.0001
-73.69
-3.96
0.0001
Table-1 and 2 shows the results of unit root test. It reveals that time series are nonstationary at levels. However, table shows that the gold price and BSE and NSE stock
price indices are stationary at 1st difference [1(1)]. Augmented Dickey Fuller unit root
analysis test reveals that errors have constant variance and are statistically independent.
Therefore, cointegration test can be applied on these variables, as supported in (Hina
Shahzadi and M.N. Chohan, 2012).
4.2 Johansen Cointegration Test Results
Consequently, Johansen cointegration test is used to determine whether there is
cointegration as well as the number of co-integrating relationships, that is, whether there
are any long-term cointegration relationships between crude oil price, gold price and BSE
SENSEX or not. Two likelihood ratio tests are used, the Trace Test and the Maximum
Eigen Value test, to determine the number of co-integrating vectors. The estimation for
each series assumes linear deterministic trend unrestricted with intercepts and no trends.
A lag of 1 to 4 (in 1st differences) is used for each series, based on the Swartz Information
Criterion (SIC).
Table-3: Johansen Cointegration Test Result
Unrestricted Cointegration Rank Test (Trace)
Hypothesized
No. of CE(s)
Trace
Eigenvalue
None *
At most 1 *
At most 2
0.0079
0.0018
0.0005
Statistic
52.96
11.67
2.34
0.05
Critical
Value
Prob.**
29.80
15.49
3.84
0.0000
0.1734
0.1259
Trace test indicates 1 cointegrating eqn(s) at the 0.05 level
* denotes rejection of the hypothesis at the 0.05 level
**MacKinnon-Haug-Michelis (1999) p-values
Unrestricted Cointegration Rank Test (Maximum Eigenvalue)
Hypothesized
Max-Eigen
No. of CE(s)
Eigenvalue
Statistic
0.05
Critical
Value
None *
At most 1 *
At most 2
0.0079
0.0018
0.0005
41.29
9.33
2.34
21.13
14.26
3.84
Prob.**
0.0000
0.2598
0.1259
Max-eigenvalue test indicates 1 cointegrating eqn(s) at the 0.05 level
* denotes rejection of the hypothesis at the 0.05 level
**MacKinnon-Haug-Michelis (1999) p-values
Included observations: 5190 after adjustments
Trend assumption: Linear deterministic trend
Lags interval (in first differences): 1 to 4
Table-3 exhibits the johansen cointegration test results. It gives surety of the long-term
relationship among the selected variables. The result shows that the series is cointegrated,
as both the trace and the maximum eigenvalue tests reject the null hypothesis of no cointegration, suggesting that there are two significant co-integrating vectors in the model.
This implies that there are two common stochastic trends, indicating a degree of market
integration. Therefore, it may conclude that there exists a stationary, long-run relationship
among the variables.
4.3 Pairwise Granger causality Tests Results
The Granger causality test is a statistical proposition test for determining whether one
time series is helpful in forecasting another. The pairwise Granger causality test has been
performed in this study in search of trend of causation among the financial variables.
Table-4 reveals that no causality exists between (i) Gold price and SENSEX, (ii) WCI
and gold price, (iii) gold price and WCI and bi-directional causality exists between (i)
SENSEX and gold price, (ii) WCI and SENSEX and (iii) SENSEX and WCI.
It is important to note that the assertion of causality between the selected variables does
not mean that movement in one variable actually causes movements in another variable.
To a certain extent, causality basically entails in order of movements in the time series
(Olushina Olawale Awe, 2012).
Table-4: Pairwise Granger Causality Test Results
Null Hypothesis
Obs
F-Statistic
Prob.
Decision
SENSEX ↑ Gold Price
5204
11.6061
9.E-06
Reject H0
Gold Price ↑ SENSEX
WCI ↑ Gold Price
Gold Price ↑ WCI
5204
5261
5261
0.81860
0.40487
2.69530
0.4411
0.6671
0.0676
DNR H0
DNR H0
DNR H0
WCI ↑ SENSEX
5196
3.04679
0.0476
Reject H0
SENSEX ↑ WCI
5196
18.6101
9.E-09
Reject H0
Type of Causality
Bi-directional
causality
No causality
No causality
No causality
Bi-directional
causality
Bi-directional
causality
Note: Decision rule: reject H0 if P-value < 0.05, DNR = Do not reject; ↑ = does not Granger cause.
5. CONCLUSIONS
This paper aspires at investigating the relationship among the crude oil price, gold price
and sensex in India. The most important conclusion of the empirical results is that the
selected time series exhibit non-stationary and that’s why providing indication of longterm cointegration relationship. Johansen cointegration test results point out that longterm cointegration stable relationships are there.
The crude oil price is a crucial impulsive variable that maneuvers as a feed throughout
which the gold price and stock prices are associated, with the target that the oil importing
countries policy makers should keep an eye on the upshots of changes in oil prices levels
on their own economies and stock markets (Mohamed Abdelaziz and Georgios
Chortareas, 2008).
During the period from 1991 to October-2012, stock markets smashed due to diverse
international crises although gold price continues to augment in India because of safe
haven financial investment as well as jewellery. World Gold Council report says that
India stands today as the world’s largest single market for gold consumption.
The measure of the relationship among the selected variables on Indian stock price
indices used in this study is based on the financial market indicators. There is necessitate
to expand this unambiguity including other macro-economic and financial market
indicators (such as interest rate, exchange rate, inflation rate, housing price) relevant to
the impact in order to arrive at more vigorous empirical analysis. This could be a possible
area for future research in India (Mishra and Mohan, 2012).
REFERENCES
Awe, Olushina Olawale. (2012). “On Pairwise Granger causality Modelling and
Econometric Analysis of Selected Economic Indicators”, [online] interstat
statjournals.net/YEAR/2012/articles/1208002.pdf, 1-17.
Dasgupta, Ranjan. 2012. Long-Run and Short-Run Relationships between
BSE SENSEX and Macroeconomic Variables, International Research Journal of Finance
and Economics, 95, 135-150.
Gilmore, Claire G, McManus, Ginette M and Sharma, Rajneesh. (2009). “The Dynamics
of Gold Prices, Gold Mining Stock Prices and Stock Market Prices Comovements”,
Research in Applied Economics, 1(1), E2, 1-19.
Hosseini, Seyed Mehdi, Ahmad, Zamri and Lai, Yew Wah. (2011). The Role of
Macroeconomic Variables on Stock Market Index in China and India, International
Journal of Economics and Finance, 3(6), 233-243.
Johansen, S., and Juselius, K. (1990). “Maximum Likelihood Estimation and
Inferences on Co-integration with Applications to the Demand for Money”, Oxford Bulletin
of Economics and Statistics, 52, 169-210.
Kaliyamoorthy, S and Parithi, S. (2012). “Relationship of Gold Market and Stock
Market: An Analysis”, International Journal of Business and Management Tomorrow,
2(6), 1-6.
Kapusuzoglu Ayhan. 2011. “Relationships between Oil Price and Stock Market: An
Empirical Analysis from Istanbul Stock Exchange (ISE)”, International Journal of
Economics and Finance, 3(6), 99-106.
Le, Thai-Ha and Chang, Youngho. (2011). “Dynamic Relationships between the Price of
Oil, Gold and Financial Variables in Japan: A Bounds Testing Approach”, Online at
http://mpra.ub.uni-muenchen.de/33030/ MPRA Paper No. 33030.
Mishra, Rabi N and Mohan, G. Jagan. (2012). “Gold Prices and Financial Stability in
India”, RBI working paper series, Department Of Economic And Policy Research, 2, 116.
Mohamed Abdelaziz, Georgios Chortareas and Andrea Cipollini. (2008). “Stock Prices,
Exchange Rates, and Oil: Evidence from Middle East Oil-Exporting Countries”, ASE
working paper, 1-27.
Nahid Kalbasi Anaraki. 2010. “The European stock market impulse to the U.S. financial
crisis, Journal of International Business and Cultural Studies”, The European Stock
Market working paper, 1-11.
Newey, W. K., & West, K. C. (1987). “A simple positive definite, heteroskedasticity and
autocorrelation consistent covariance matrix”. Econometrica, 55, 703-708.
Osterwald-Lenum, M. (1992). “A Note with Quintiles of the Asymptotic Distribution of
the Maximum Likelihood Co-integration Rank Test Statistic”, Oxford Bulletin of
Economics and Statistics, 54, 461-472.
Rad, Abbas Alavi. 2011. “Macroeconomic Variables and Stock Market: Evidence from
Iran”, International Journal of Economics and Finance Studies, 3(1), 1-10.
Samanta, Subarna K. and Zadeh, Ali H. M. (2012). “Co-Movements of Oil, Gold, the US
Dollar, and Stocks”, Modern Economy, 3, 111-117.
Shahzadi, Hina and Chohan, Muhammad Naveed. (2012). “Impact of Gold Prices on
Stock Exchange: A Case Study of Pakistan”, Working paper series, Karachi Stock
Exchange, 10 (2), 1-12.
Sharma, Vivek. (2012). “Three critical economic factors that influence the Indian stock
market”, Money life, [online] http://www.moneylife.in/article/three-critical-economicfactors-that-influence-the-indian-stock-market/28695.html.
Sharma, Gagan Deep and Mahendra, Mandeep. (2010). “Impact of Macro-economic
Variables on Stock Prices in India”, Global Journal of Management and Business
Research, 10(7), 19-26.
Šimáková, Jana, 2011. “Analysis of the Relationship between Oil and Gold Prices”,
Journal of Finance, 51(1), 651-662.
Thai-Ha Le1 and Youngho Chang (2011). Oil And Gold: Correlation or Causation?,
Depocen, Working Paper Series No. 2011/22, 1-36.
Wang Mu-Lan, Wang Ching-Ping and Huang Tzu-Ying. (2010). “Relationships among
Oil Price, Gold Price, Exchange Rate and International Stock Markets”, International
Research Journal of Finance and Economics, 47, 80-89.
Yahyazadehfar, Mahmood and Babaie, Ahmad. (2012). “Macroeconomic Variables and
Stock Price: New Evidence from Iran”, Middle-East Journal of Scientific Research, 11
(4), 408-415.